Information processing device, information processing method, and recording medium

Abstract

[Problem] Conventionally, it has not been easy to utilize generative AI by using an appropriate prompt according to the class of inquiry information. [Solution] An information processing device 1 comprises: a template management unit 111 in which prompt template information is stored in association with one or more class conditions of a character string; an inquiry acquisition unit 131 that acquires, from reception information, inquiry information, which is used for making an inquiry to generative AI; a template acquisition unit 134 that acquires, from the template management unit 111, prompt template information corresponding to a class condition satisfied by the class of the inquiry information; a prompt creation unit 135 that creates a prompt by using the prompt template information and the inquiry information; a response acquisition unit 136 that gives the prompt to the generative AI and acquires a response from the generative AI; an output acquisition unit 137 that uses the response to acquire output information; and an output unit 14 that outputs the output information. Through the information processing device 1, generative AI can be utilized by using an appropriate prompt according to the class of the inquiry information.

Description

Information processing device, information processing method, and recording medium

[0001] The present invention relates to an information processing device that acquires and outputs information using generation AI.

[0002] Recently, generative AI has become popular (see Patent Document 1).

[0003] “ChatGPT”, [online], [searched November 19, 2023], Internet [URL: https: / / openai.com / chatgpt]

[0004] However, in the prior art, it was not easy to provide appropriate prompts to the generation AI. More specifically, it was not easy to utilize the generation AI with appropriate prompts according to the class of query information. Note that the generation AI here is an AI that generates text.

[0005] The information processing device of the first invention is an information processing device comprising: a template management unit in which prompt template information is stored in correspondence with one or more class conditions, which are conditions related to a class of strings; a reception unit for receiving reception information from a user; a query acquisition unit for acquiring query information, which is information contained in the reception information and is information used to query the generation AI; a template acquisition unit for acquiring prompt template information from the template management unit corresponding to the class conditions satisfied by the class of the query information acquired by the query acquisition unit; a prompt creation unit for creating a prompt using the prompt template information acquired by the template acquisition unit and the query information acquired by the query acquisition unit; an answer acquisition unit for passing the prompt created by the prompt creation unit to the generation AI and acquiring an answer from the generation AI; an output acquisition unit for acquiring output information using the answer acquired by the answer acquisition unit; and an output unit for outputting the output information.

[0006] This configuration allows the generation AI to be used with appropriate prompts depending on the class of query information.

[0007] In addition, the information processing device of the second invention is an information processing device in which, compared to the first invention, the received information is information related to the invention and the output information is information constituting a patent specification, claims, or abstract.

[0008] This configuration allows generative AI to be used to assist in the creation of patent documents with appropriate prompts depending on the class of invention information.

[0009] Furthermore, the information processing device of the third invention is an information processing device in which, compared to the first or second invention, the inquiry acquisition unit acquires at least two pieces of inquiry information from the reception information, and the prompt creation unit creates a prompt using each of the two pieces of inquiry information.

[0010] This configuration allows the generation AI to be used with appropriate prompts depending on the class of query information.

[0011] Furthermore, the information processing device of the fourth invention is an information processing device in which, compared to any one of the first to third inventions, the output acquisition unit places part or all of the answer in output template information, which is a template for constructing output information, and acquires the output information.

[0012] With this configuration, appropriate output information can be obtained using the generation AI with an appropriate prompt according to the class of the query information.

[0013] Furthermore, in the information processing device of the fifth invention, in relation to any one of the first to fourth inventions, the class is an information processing device that is a linguistic characteristic of the query information.

[0014] This configuration allows the appropriate class of query information to be determined and the generating AI to be used with appropriate prompts depending on that class.

[0015] Furthermore, the information processing device of the sixth invention is an information processing device in which, compared to the fifth invention, the class can be "question", the template acquisition unit does not acquire prompt template information when the class determined by the class determination unit is "question", and the prompt creation unit creates a prompt using the query information acquired by the query acquisition unit.

[0016] This configuration allows the appropriate class of query information to be determined and the generating AI to be used with appropriate prompts depending on that class.

[0017] Furthermore, an information processing device according to a seventh aspect of the present invention is the information processing device according to any one of the first to sixth aspects of the present invention, wherein the class is a tag included in the reception information and the inquiry information is a corresponding tag.

[0018] This configuration allows the appropriate class of query information to be determined and the generating AI to be used with appropriate prompts depending on that class.

[0019] An information processing device according to an eighth aspect of the present invention is the information processing device according to any one of the first to seventh aspects of the present invention, wherein the received information is a document and the class is a document type.

[0020] This configuration allows the appropriate class of query information to be determined and the generating AI to be used with appropriate prompts depending on that class.

[0021] In addition, the information processing device of the ninth invention is an information processing device described in any one of the first to eighth inventions, further comprising a generation AI determination unit that determines a generation AI from two or more generation AIs according to the prompt template information acquired by the template acquisition unit, and the answer acquisition unit passes the prompt created by the prompt creation unit to the generation AI determined by the generation AI determination unit and acquires an answer from the generation AI.

[0022] This configuration makes it possible to provide a generative AI control engine that makes the most of generative AI.

[0023] According to the information processing device of the present invention, the generation AI can be utilized using an appropriate prompt according to the class of the inquiry information.

[0024] FIG. 1 is a conceptual diagram of an information system A in the first embodiment; FIG. 2 is a block diagram of the information system A; FIG. 3 is a flowchart illustrating an example of the operation of the information processing device 1; FIG. 4 is a flowchart illustrating an example of the received information processing; FIG. 5 is a flowchart illustrating an example of the class condition processing; FIG. 6 is a flowchart illustrating an example of the output acquisition processing; FIG. 7 is a diagram showing the template management table; FIG. 8 is a diagram showing an example of the received information; FIG. 9 is a diagram showing an example of the answer; FIG. 10 is a diagram showing an example of the received information;

[0025] Hereinafter, embodiments of an information processing device and the like will be described with reference to the drawings. Note that components with the same reference numerals in the embodiments perform similar operations, and therefore repeated description may be omitted.

[0026] (Embodiment 1) In this embodiment, an information processing device is described that determines the class of inquiry information, obtains prompt template information corresponding to the class, creates a prompt using the prompt template information, provides the prompt to a generation AI, obtains an answer, and outputs output information using the answer. In particular, an information processing device is described that uses inquiry information related to an invention to output a portion of a patent document. The class is, for example, information based on the linguistic characteristics of a string, a tag corresponding to the string, or the type of document accepted. Note that a prompt is a question provided to the generation AI.

[0027] In this embodiment, an information processing device will be described that acquires two or more pieces of inquiry information from the reception information and creates a prompt using each of the two or more pieces of inquiry information.

[0028] In addition, in this embodiment, an information processing device that provides an answer acquired from the generation AI to output template information and acquires output information will be described.

[0029] Furthermore, in this embodiment, an information processing device will be described that determines a generation AI according to a class and acquires output information using a response from the generation AI.

[0030] In this specification, information X being associated with information Y means that information Y can be obtained from information X, or information X can be obtained from information Y, and the method of association is not important. Information X and information Y may be linked, may exist in the same buffer, information X may be included in information Y, or information Y may be included in information X, etc.

[0031] Furthermore, in this specification, selecting or determining information Z means obtaining information Z, obtaining a pointer to information Z, obtaining the ID of information Z, setting a flag on information Z, etc., and it is sufficient if information Z can be accessed.

[0032] 1 is a conceptual diagram of an information system A according to this embodiment. The information system A includes an information processing device 1, one or more terminal devices 2, and one or more generation AI devices 3.

[0033] The information processing device 1 is a device that receives reception information, obtains an answer using a generation AI, and outputs output information using the answer. The information processing device 1 is, for example, a cloud server or an ASP server, but the type does not matter. The information processing device 1 may also be a terminal. If the information processing device 1 is a terminal, it can be considered that the terminal device 2 is not necessary in the information system A, or that the information processing device 1 also serves as the terminal device 2. The information processing device 1 may include a generation AI.

[0034] The terminal device 2 is a terminal used by a user to input reception information. The terminal device 2 may be, for example, a personal computer, a smartphone, or a tablet terminal, but the type of the terminal device 2 is not important.

[0035] The generating AI device 3 is a device having a generating AI that receives a prompt and outputs an answer. If the information processing device 1 includes a generating AI, the generating AI device 3 may be omitted.

[0036] 2 is a block diagram of an information system A according to this embodiment. The information processing device 1 includes a storage unit 11, a reception unit 12, a processing unit 13, and an output unit 14. The storage unit 11 includes a template management unit 111. The processing unit 13 includes a query acquisition unit 131, a class determination unit 132, a generation AI determination unit 133, a template acquisition unit 134, a prompt creation unit 135, an answer acquisition unit 136, and an output acquisition unit 137.

[0037] The terminal device 2 includes a terminal storage unit 21 , a terminal reception unit 22 , a terminal processing unit 23 , a terminal transmission unit 24 , a terminal reception unit 25 , and a terminal output unit 26 .

[0038] Various types of information are stored in the storage unit 11 that constitutes the information processing device 1. The various types of information include, for example, prompt template information, which will be described later.

[0039] Two or more pieces of prompt template information are stored in the template management unit 111. Each of the two or more pieces of prompt template information is associated with a class condition.

[0040] Prompt template information is information that serves as a template for creating a prompt. Prompt template information is usually a character string. Prompt template information has one or more variables. The variables are variables into which query information, which will be described later, is substituted. Prompt template information, for example, has a character string that specifies a question and one or more variables. Note that, for example, query information is placed in the variable portion of the prompt template information, and a prompt is created.

[0041] The template management unit 111 may store output template information. Output template information is information that serves as a template for creating output information. Output template information is usually a character string. The output template information has one or more variables. The variables are variables into which part or all of the answers described below are assigned. For example, an answer or part of an answer is placed in the place of the variable in the output template information, and output information is created.

[0042] A class condition is a condition related to a class, which will be described later. A class condition is a condition related to one or more classes. A class relates to a character string. A class may also be called a classification or type. A class may also be a field identifier. A class may be identified by one or more field identifiers. A field identifier is information that identifies a field on an input screen, which is a screen for inputting information. A field identifier is, for example, a field name or a field ID. Note that a field is broadly understood to refer to a space or element where a user inputs, displays, or selects information. A field may be considered to include, for example, a text box for inputting "name" or "address" on a form screen, a drop-down menu, a check box, a radio button, a date field, etc.

[0043] The reception unit 12 receives reception information. The reception unit 12 typically receives reception information from a user. The reception information from a user is reception information provided by the user. The reception unit 12 typically receives reception information from the terminal device 2.

[0044] Here, acceptance typically refers to the reception of information transmitted via a wired or wireless communication line, but may also be a concept that includes the reception of information input from an input device such as a keyboard, mouse, or touch panel, or the reception of information read from a recording medium such as an optical disk, magnetic disk, or semiconductor memory.

[0045] The received information is usually information entered by a user. However, the received information may also be information obtained from a file or a database. The received information is, for example, information related to an invention. The received information is, for example, information that identifies an invention. The received information is, for example, claims, a specification, or an abstract. The received information is, for example, a document. The received information is, for example, a file. The received information is, for example, a sentence, one or more sentences, or a character string. It goes without saying that the received information may also be information entered by a user on an input screen having one or more fields. It goes without saying that the received information may also be information entered into one or more fields.

[0046] The processing unit 13 performs various types of processing, such as processing performed by the inquiry acquisition unit 131, the class determination unit 132, the generation AI determination unit 133, the template acquisition unit 134, the prompt creation unit 135, the answer acquisition unit 136, or the output acquisition unit 137.

[0047] The inquiry acquisition unit 131 acquires inquiry information from the reception information received by the reception unit 12. The inquiry information is information used to inquire of the generation AI. The inquiry information is information included in the reception information. The inquiry information may be reception information. The inquiry information is a character string. The inquiry information is, for example, one or more sentences, words, or words of a specific part of speech (for example, a noun).

[0048] The inquiry acquisition unit 131 acquires, for example, at least two pieces of inquiry information from the reception information received by the reception unit 12 .

[0049] The one or more pieces of inquiry information that the inquiry acquisition unit 131 acquires from the reception information are usually predetermined.

[0050] The inquiry acquisition unit 131 acquires, for example, one or more pieces of inquiry information classified into a class that matches the class condition from the received information.

[0051] The inquiry acquisition unit 131 may acquire inquiry information based on information entered in one or more fields. The inquiry information may be a part or all of the information entered in the fields. The inquiry information may be a part or all of the information entered in two or more fields.

[0052] The class determination unit 132 determines the class of the inquiry information. The class determination unit 132 may also determine the class of the received information. For example, the class determination unit 132 determines the class of each of the one or more pieces of inquiry information acquired by the inquiry acquisition unit 131.

[0053] The class may also be referred to as a type, a classification, etc. The class is, for example, information based on linguistic characteristics, information based on a tag to which the query information corresponds, or information based on the type of document containing the query information.

[0054] Linguistic characteristics may be referred to as linguistic attribute values ​​for character strings. A class based on linguistic characteristics is, for example, "question" or "non-question." A class based on linguistic characteristics is, for example, "sentence" or "word." A class based on linguistic characteristics is, for example, "sentence," "paragraph," or "word." A class based on linguistic characteristics is, for example, the part of speech of a word. A class based on linguistic characteristics is, for example, "component" or "technical term" that identifies the invention. "Technical term" may also be "technical term."

[0055] The tags to which the inquiry information corresponds are, for example, tags in the specification, tags in the scope of the claims, and tags in the abstract. Classes based on tags are, for example, "Problem to be Solved by the Invention," "Means for Solving the Problem," "Effects of the Invention," "Form for Carrying Out the Invention," "Explanation of Symbols," and "Claim 1." The tags to which the inquiry information corresponds are, for example, HTML tags or XML tags. The tags may be one or more field identifiers. In other words, it goes without saying that the inquiry information may be information entered in a feed identified by one or more field identifiers.

[0056] A class based on a document type is a class based on the type of reception information, which is a file. Examples of classes based on a document type include "specification," "claims," ​​"abstract," "invention list," and "sales report."

[0057] The class determination unit 132, for example, acquires linguistic characteristics of the inquiry information and determines a class corresponding to the linguistic characteristics.

[0058] The class determination unit 132 acquires, from the reception information, a tag that is included in the reception information and to which the inquiry information corresponds, and determines a class that corresponds to the tag.

[0059] The class determination unit 132 acquires, for example, the type of document that is the reception information, and determines the class corresponding to that type.

[0060] The generation AI determination unit 133 determines a generation AI according to a class, for example, from two or more generation AIs. In such a case, each of the two or more generation AIs is associated with a class condition. The generation AI determination unit 133 determines, for example, a generation AI that pairs with a class condition. Determining a generation AI means, for example, obtaining an API for using the generation AI. In such a case, the class condition is associated with, for example, an API for passing a prompt to the generation AI. Determining a generation AI means, for example, obtaining an identifier (e.g., IP address) of the generation AI device 3 that stores the generation AI. In such a case, the class condition is associated with the identifier of the generation AI device 3.

[0061] The template acquisition unit 134 acquires prompt template information corresponding to a class from the template management unit 111. The template acquisition unit 134 acquires prompt template information corresponding to a class condition from the template management unit 111, for example.

[0062] It is preferable that the template acquiring unit 134 does not acquire prompt template information when the class of the received reception information is "question."

[0063] The prompt creation unit 135 creates a prompt using the prompt template information acquired by the template acquisition unit 134 and the query information acquired by the query acquisition unit 131. The prompt creation unit 135 typically places the acquired query information in the variable portion of the prompt template information acquired by the template acquisition unit 134 to create a prompt.

[0064] The prompt generator 135 may generate a prompt using each of the two pieces of inquiry information. The number of prompts generated here may be one or two.

[0065] If the class of the inquiry information acquired by the inquiry acquisition unit 131 is a "question," the prompt creation unit 135 creates a prompt including the inquiry information. The prompt may be the inquiry information.

[0066] The answer acquisition unit 136 passes the prompt created by the prompt creation unit 135 to the generation AI and acquires an answer from the generation AI. Passing to the generation AI means, for example, passing to the generation AI device 3, but it may also be passing to the generation AI possessed by the information processing device 1.

[0067] The answer acquisition unit 136 passes the prompt created by the prompt creation unit 135 to the generation AI, for example, using the API of the generation AI.

[0068] It is preferable that the answer acquisition unit 136 passes the prompt created by the prompt creation unit 135 to the generation AI determined by the generation AI determination unit 133 and acquires an answer from the generation AI.

[0069] The output acquisition unit 137 acquires output information using the answer acquired by the answer acquisition unit 136. The output acquisition unit 137 acquires output information using, for example, the reception information and the answer acquired by the answer acquisition unit 136. The output acquisition unit 137, for example, places the answer in the reception information and acquires the output information. The output acquisition unit 137, for example, replaces part or all of the reception information with the answer and acquires the output information. The output acquisition unit 137 may acquire the output information using the answer from the generation AI and the processing type. The output acquisition unit 137 may acquire the output information to be executed by the output unit 14. In such a case, the output information is, for example, a function, a method, or an execution module.

[0070] It is preferable that the output acquisition unit 137 acquires the output information by placing part or all of the answer in the variable portion of the output template information.

[0071] The output information is, for example, information that constitutes a patent specification, claims, or abstract.

[0072] The output unit 14 outputs the output information acquired by the output acquisition unit 137. The output unit 14, for example, transmits the output information to the terminal device 2. The output unit 14 may perform processing using the output information acquired by the output acquisition unit 137. The output unit 14 may execute a function, a method, or an execution module.

[0073] Here, output usually means transmission to terminal device 2, but it may also be a concept that includes display on a display, projection using a projector, printing on a printer, sound output, transmission to other external devices, storage on a recording medium, and handing over processing results to other processing devices or other programs.

[0074] Various types of information are stored in the terminal storage unit 21 of the terminal device 2. The various types of information include, for example, reception information and a user identifier.

[0075] The terminal reception unit 22 receives various types of information, instructions, etc. The various types of information, instructions, etc. are, for example, reception information.

[0076] The means for inputting various information and instructions may be any means, such as a touch panel, keyboard, mouse, or menu screen.

[0077] The terminal processing unit 23 performs various types of processing. For example, the various types of processing are processing to convert received information, instructions, etc. into information, instructions, etc. with a structure to be transmitted. For example, the various types of processing are processing to convert received information into information with a structure to be output.

[0078] The terminal transmitting unit 24 transmits various types of information, instructions, etc. to the information processing device 1. The various types of information, instructions, etc. are, for example, reception information.

[0079] The terminal receiving unit 25 receives various types of information from the information processing device 1. The various types of information are, for example, output information.

[0080] The terminal output unit 26 outputs various types of information, such as output information.

[0081] The storage unit 11, template management unit 111, and terminal storage unit 21 are preferably non-volatile recording media, but may also be realized as volatile recording media.

[0082] There is no restriction on the process by which information is stored in the storage unit 11 etc. For example, information may be stored in the storage unit 11 etc. via a recording medium, information transmitted via a communication line etc. may be stored in the storage unit 11 etc., or information input via an input device may be stored in the storage unit 11 etc.

[0083] The reception unit 12 is preferably realized by a wireless or wired communication means, but may also be realized by a means for receiving broadcasts, a device driver for an input means such as a touch panel or keyboard, or control software for a menu screen.

[0084] The processing unit 13, inquiry acquisition unit 131, class determination unit 132, generation AI determination unit 133, template acquisition unit 134, prompt creation unit 135, answer acquisition unit 136, output acquisition unit 137, and terminal processing unit 23 can typically be realized by a processor, memory, etc. The processing procedures of the processing unit 13, etc. are typically realized by software, and the software is recorded on a recording medium such as a ROM. However, they may also be realized by hardware (dedicated circuit). The processor may be a CPU, MPU, GPU, etc., and the type does not matter.

[0085] The output unit 14 is usually realized by a wireless or wired communication means, but may also be realized by a broadcasting means. The output unit 14 may also be realized by driver software for an output device such as a display or speaker, or by a combination of driver software for an output device and the output device.

[0086] The terminal reception unit 22 can be realized by a device driver for an input means such as a touch panel or a keyboard, or control software for a menu screen.

[0087] The terminal transmitting unit 24 is usually realized by a wireless or wired communication means, but may also be realized by a broadcasting means.

[0088] The terminal receiving unit 25 is usually realized by a wireless or wired communication means, but may also be realized by a means for receiving broadcasts.

[0089] The terminal output unit 26 may or may not be considered to include an output device such as a display, a speaker, etc. The terminal output unit 26 may be realized by driver software for an output device, or by a combination of driver software for an output device and the output device, etc.

[0090] Next, an example of the operation of the information processing device 1 will be described with reference to the flowchart of FIG.

[0091] (Step S301) The reception unit 12 determines whether or not reception information has been received from the user. If reception information has been received, the process proceeds to step S302. If reception information has not been received, the process returns to step S301. Here, the reception unit 12 determines whether or not reception information has been received from the terminal device 2, for example.

[0092] (Step S302) The inquiry acquisition unit 131 determines whether the class of the reception information received in step S301 is a "question" or a "non-question." If it is a "question," the process proceeds to step S303; if it is not a "question," the process proceeds to step S308. Note that the entire reception information is a "question" when it consists of a predetermined character string, such as "Is it...?" or "Is it...?". Furthermore, whether a character string is a "question" or a "non-question" can be determined by machine learning prediction processing (publicly known technology).

[0093] (Step S303) The prompt generator 135 generates a prompt including the reception information received in step S301. The prompt is, for example, reception information.

[0094] (Step S304) The answer acquisition unit 136 passes the prompt constructed in step S303 to the generation AI. Here, the answer acquisition unit 136, for example, assigns the prompt to an argument of the API of the generation AI and executes the API.

[0095] (Step S305) The answer acquisition unit 136 determines whether or not an answer has been acquired from the generated AI. If an answer has been acquired, the process proceeds to step S306. If an answer has not been acquired, the process returns to step S305. Here, the answer acquisition unit 136 acquires the answer, which is, for example, the return value of the executed API.

[0096] (Step S306) The output acquisition unit 137 generates output information including all or part of the answer acquired in step S305. When the output acquisition unit 137 acquires part of the answer, for example, it manages information for determining the part to acquire (for example, "information specifying the part of the text") and acquires the part of the answer using such information.

[0097] (Step S307) The output unit 14 outputs the output information constructed in step S306. The process returns to step S301.

[0098] (Step S308) The inquiry acquisition unit 131 etc. processes the reception information received in step S301 and outputs output information. Return to step S301. An example of such reception information processing will be described with reference to the flowchart of FIG.

[0099] In the flowchart of FIG. 3, the process ends when the power is turned off or an interrupt occurs to end the process.

[0100] Next, an example of the reception information processing in step S308 will be described with reference to the flowchart of FIG.

[0101] (Step S401) The class determination unit 132 acquires the document type, which is the reception information. Note that here, the class determination unit 132 may not be able to acquire the document type. The document type may be a "class." The class determination unit 132 determines the document type, for example, from the file name of the reception information. The class determination unit 132 determines the document type, for example, from a character string corresponding to a specific tag (e.g., [document name]) in the reception information.

[0102] The class determination unit 132, for example, provides the reception information and the learning model to a machine learning prediction module, executes the prediction module, and obtains the document type. The learning model is information configured by a machine learning learning process and is information used in the machine learning prediction process. The learning model may also be referred to as a learner, a classifier, a classification model, etc. In this specification, the machine learning algorithm may be deep learning, a random forest, a decision tree, an SVM, etc. Furthermore, for machine learning, various machine learning functions such as the TensorFlow (registered trademark) library, the random forest module of the R language, fastText, TinySVM, etc., or various existing libraries may be used.

[0103] (Step S402) The class determination unit 132 obtains one or more class conditions that match the type obtained in step S401 from the template management unit 111. If the document type cannot be obtained in step S401, the class determination unit 132 obtains one or more class conditions that do not correspond to any document type from the template management unit 111.

[0104] (Step S403) The class determination unit 132 assigns 1 to a counter i.

[0105] (Step S404) The class determination unit 132 determines whether the i-th class condition exists among the class conditions acquired in step S402. If the i-th class condition exists, the process proceeds to step S405; if not, the process returns to the upper level process.

[0106] (Step S405) The processing unit 13 processes the reception information using the i-th class condition. An example of such class condition processing will be described with reference to the flowchart of FIG.

[0107] (Step S406) The class determination unit 132 increments the counter i by 1. The process returns to step S404.

[0108] Next, an example of the class condition processing in step S405 will be described with reference to the flowchart of FIG.

[0109] (Step S501) The template acquisition unit 134 acquires, from the template management unit 111, prompt template information paired with the i-th class condition of step S404.

[0110] (Step S502) The query acquisition unit 131 determines whether or not a tag exists in the i-th class condition. If a tag exists, the process proceeds to step S503; if not, the process proceeds to step S505. Note that the presence of a tag in a class condition means that the tag is a condition.

[0111] (Step S503) The query acquisition unit 131 acquires a tag in the i-th class condition.

[0112] (Step S504) The inquiry acquiring unit 131 acquires, from the received information, tag content, which is information corresponding to the tag acquired in step S503. The tag content is usually a character string corresponding to the tag.

[0113] (Step S505) The query acquisition unit 131 determines whether or not a linguistic feature is present in the i-th class condition. If a linguistic feature is present, the process proceeds to step S504; if not, the process proceeds to step S508. Note that the presence of a linguistic feature in a class condition means that the linguistic feature is a condition.

[0114] (Step S506) The query acquisition unit 131 acquires the linguistic characteristics of the i-th class condition.

[0115] (Step S507) If the query acquisition unit 131 acquires a tag in the i-th class condition, it acquires one or more pieces of query information that are information in the tag content acquired in step S504 and match the linguistic characteristics acquired in step S506. If the query acquisition unit 131 does not acquire a tag in the i-th class condition, it acquires one or more pieces of query information that are information in the received reception information and match the linguistic characteristics acquired in step S506. Then, it proceeds to step S509.

[0116] (Step S508) If a tag exists in the i-th class condition, the query acquisition unit 131 acquires the tag content acquired in step S504 as query information. If a tag does not exist in the i-th class condition, the query acquisition unit 131 acquires the received reception information as query information.

[0117] (Step S509) The prompt generator 135 assigns 1 to a counter i.

[0118] (Step S510) The prompt generator 135 determines whether the i-th inquiry exists among the inquiries acquired in step S507 or step S508. If the i-th inquiry exists, the process proceeds to step S510; if not, the process returns to the upper level process.

[0119] (Step S511) The prompt generator 135 acquires the i-th inquiry information from the inquiry information acquired in step S507 or step S508. The prompt generator 135 places the i-th inquiry information in the variable portion of the prompt template information acquired in step S501, and creates a prompt.

[0120] (Step S512) The answer acquisition unit 136 passes the prompt created in step S511 to the generation AI.

[0121] (Step S513) The answer acquisition unit 136 determines whether an answer has been acquired from the generation AI. If an answer has been acquired, the process proceeds to step S514; if an answer has not been acquired, the process returns to step S513.

[0122] (Step S514) The output acquisition unit 137 acquires the output information. An example of such output acquisition processing will be described with reference to the flowchart of FIG.

[0123] (Step S515) The prompt generator 135 increments the counter i by 1. The process returns to step S510.

[0124] Next, an example of the output acquisition process in step S514 will be described with reference to the flowchart in FIG.

[0125] (Step S601) The output acquisition unit 137 acquires a process type paired with a class condition from the template management unit 111.

[0126] (Step S602) The output acquisition unit 137 determines whether the processing type acquired in step S601 is “addition.” If it is “addition,” the process proceeds to step S603; if it is not “addition,” the process proceeds to step S604.

[0127] (Step S603) The output acquisition unit 137 places the acquired answer or part of the answer in a predetermined position for the inquiry information, and returns to the upper-level processing.

[0128] (Step S604) The output acquisition unit 137 determines whether the processing type acquired in step S601 is “change.” If it is “change,” the process proceeds to step S605, and if it is not “change,” the process proceeds to step S606.

[0129] (Step S605) The output acquisition unit 137 updates the inquiry information to the answer or part of the answer, and returns to the upper-level process.

[0130] (Step S606) The output acquisition unit 137 determines whether the processing type acquired in step S601 is "check." If it is "check," the process proceeds to step S607; if it is not "check," the process returns to the upper processing. Note that "check" may also be called "inspection."

[0131] (Step S607) The output acquisition unit 137 associates the answer or part of the answer with the inquiry information, and returns to the upper-level process.

[0132] A specific example of the operation of the information system A in this embodiment will be described below.

[0133] Currently, the template management table shown in Figure 7 is stored in the template management unit 111 of the information processing device 1. The template management table has "ID," "class condition," "prompt template information," and "processing type." The "class condition" has "tag" and "linguistic characteristic." Note that an identifier of the generation AI device 3 is associated with the "prompt template information," and an answer may be obtained using the generation AI device 3 that is paired with the prompt template information to be used.

[0134] In FIG. 7, "ID" is information that identifies a record. "Tag" is a tag that can be included in the reception information. An example of a tag included in the invention list is [Invention 1]. An example of a tag included in the specification is [Form for implementing the invention]. "Linguistic characteristics" are, for example, "sentence," "paragraph," "component," and "technical term." In the "prompt template information," <inquiry information> is a variable, and <inquiry information> is replaced with the inquiry information acquired by the inquiry acquisition unit 131.

[0135] "Processing type" is information indicating the type of processing when adding an answer. Here, "processing type" is information indicating the type of processing when adding an answer to reception information. Here, "processing type" is "add," "check," or "change." Also, here, the attribute values ​​of the string of the output answer differ depending on the "processing type." The attribute values ​​of the string are, for example, font, color, size, and whether or not it is underlined. "Processing type" may also be information that identifies the processing using the answer. The processing type "insert (transportation expenses, expense DB)" means performing a process to register the transportation expenses obtained from the generation AI in the expense DB. The processing type "send (summary, $supervisor address)" means sending the summary (summary of the sales report) obtained from the generation AI by email to the address indicated by "$supervisor address." Note that "$Supervisor's Address" is, for example, a variable that indicates the email address of the user's supervisor that is paired with the user identifier of the user who entered the sales daily report, and is information obtained from an address database (not shown) that manages one or more pairs of user identifiers and supervisor's email addresses.

[0136] In Figure 7, if the processing type is "add", it indicates that the output acquisition unit 137 adds an answer to the received information. If the processing type is "add", it indicates that the output acquisition unit 137 adds an answer immediately after a sentence having inquiry information (including a sentence that is inquiry information). Also, if the processing type is "check", it indicates that the output acquisition unit 137 adds an answer to the received information in association with the inquiry information to be checked. If the processing type is "change", it indicates that the output acquisition unit 137 replaces the inquiry information in the received information with an answer.

[0137] In the above situation, four specific examples will be explained. Specific Example 1 is when the document is an invention list. Specific Example 2 is when the document is a specification. Specific Example 3 is when the document is an abstract. Specific Example 4 is when the document is a sales daily report entered on a screen.

[0138] (Specific Example 1) It is assumed that the user inputs into the information processing device 1 a file "invention list 20231125.docx" in which the character string shown in FIG.

[0139] Next, the reception unit 12 of the information processing device 1 receives reception information "Invention list 20231125.docx" from the user. Next, the inquiry acquisition unit 131 opens the file "Invention list 20231125.docx", acquires its contents, detects that it does not end with "?", and determines that the received reception information is not a "question" (is a "non-question").

[0140] Next, the inquiry acquisition unit 131 etc. processes the received reception information as follows, and outputs output information.

[0141] First, the class determination unit 132 obtains the document type "invention list" from the file name of the reception information "invention list 20231125.docx".

[0142] Next, the class determination unit 132 acquires the class conditions for "ID=1" and "ID=4" that match the type "invention list." Note that the document type "*" for "ID=4" indicates that it matches any document type.

[0143] Next, the class determination unit 132 acquires the class condition "<tag> invention 1 <linguistic property>-" for "ID=1" from the template management table (FIG. 7).

[0144] The template acquisition unit 134 also acquires the prompt template information "Tell me the subconcept of the following invention! Invention "<Inquiry information>"" that is paired with the class condition of "ID=1" from the template management table (FIG. 7).

[0145] Next, the query acquisition unit 131 acquires the tag "Invention 1" in the class condition from the template management table (FIG. 7). Next, the query acquisition unit 131 acquires the tag content "electrical device that detects the presence of a person and changes the advertisement output," which is information corresponding to the acquired tag "Invention 1," from the reception information (FIG. 8). Next, the query acquisition unit 131 determines that no linguistic characteristics exist in the class condition. Next, the query acquisition unit 131 sets the acquired tag content "electrical device that detects the presence of a person and changes the advertisement output" as query information.

[0146] Next, the prompt creation unit 135 replaces the variable <inquiry information> of the acquired prompt template information with "an electrical device that detects the presence of a person and changes the output of an advertisement" and acquires the prompt "Please tell me the sub-concept of the following invention! Invention "an electrical device that detects the presence of a person and changes the output of an advertisement"".

[0147] Next, the answer acquisition unit 136 passes the created prompt to the generation AI. Then, the answer acquisition unit 136 acquires the answer from the generation AI. Note that it is assumed that the answer is, for example, the sentence 901 in FIG. 9 .

[0148] Next, the output acquisition unit 137 acquires output information as follows: That is, the output acquisition unit 137 acquires the process type "add" that is paired with the class condition of "ID=1" from the template management table.

[0149] Next, the output acquisition unit 137 places the acquired response to the inquiry information at a predetermined position (here, immediately after the tag sentence (inquiry information) of Invention 1) and obtains the information shown in FIG.

[0150] Next, the class determination unit 132 acquires the class condition "<tag>-<linguistic property> statement" for "ID=4" from the template management table (FIG. 7). The information processing device 1 then performs the same process as above for the class condition for "ID=4", but in this case, the output acquisition unit 137 does not acquire output information.

[0151] (Specific Example 2) It is assumed that the user inputs into the information processing device 1 a file "specification 20231125.docx" that stores the character string shown in FIG.

[0152] Next, the reception unit 12 of the information processing device 1 receives the reception information "specification20231125.docx" from the user. Next, the inquiry acquisition unit 131 determines that the reception information is not a "question" based on the contents of the received file "specification20231125.docx."

[0153] Next, the inquiry acquisition unit 131 etc. processes the received reception information as follows, and outputs output information.

[0154] First, the class determination unit 132 obtains the character string "specification" paired with the tag [document name] in the file of reception information "specification20231125.docx", and obtains the document type "specification" from the character string.

[0155] Next, the class determination unit 132 acquires the class conditions for "ID=2" and "ID=4" that match the type "specification."

[0156] Next, the class determination unit 132 acquires the class conditions "<tag>-<linguistic property> sentence," "<tag>-<linguistic property> paragraph," "<tag> mode for carrying out the invention <linguistic property> component," and "<tag> mode for carrying out the invention <linguistic property> technical term" for "ID=2" from the template management table (FIG. 7). The class determination unit 132 also acquires the class condition "<tag>-<linguistic property> sentence" for "ID=4" from the template management table (FIG. 7).

[0157] The template acquisition unit 134 also acquires prompt template information paired with each class condition of "ID=2" from the template management table (FIG. 7) in association with each class condition.

[0158] Next, the query acquisition unit 131 sequentially acquires sentences from the file “specification20231125.docx” based on the class condition “<tag>-<linguistic property> sentence” for “ID=2.” Each sentence is query information.

[0159] Next, the prompt creation unit 135 places each sentence in the <inquiry information> section of the prompt template information "Please tell us, YES / NO, whether the following sentence has a subject or not. Sentence "<inquiry information>". As a result, the prompt creation unit 135 creates, for example, a prompt "Please tell us, YES / NO, whether the following sentence has a subject or not. Sentence "The inquiry acquisition unit 131 acquires inquiry information, which is information contained in the reception information and is information used to query the generation AI." and a prompt "Please tell us, YES / NO, whether the following sentence has a subject or not. Sentence "At least two pieces of inquiry information are acquired from the reception information."

[0160] Next, the answer acquisition unit 136 passes each created prompt to the generation AI. Then, the answer acquisition unit 136 acquires an answer ("YES" or "NO") from the generation AI for each prompt. Note that the answer to the inquiry information "Acquire inquiry information that is information included in the reception information and is information to be used to inquire to the generation AI" was "YES", and the answer to the inquiry information "Acquire at least two pieces of inquiry information from the reception information" was "NO".

[0161] Furthermore, the query acquisition unit 131 sequentially acquires all of the sentences in the paragraphs corresponding to each paragraph number from the file "Specification20231125.docx" based on the class condition "<tag>-<linguistic characteristic> paragraph" for "ID=2." All of the sentences in the paragraphs are query information.

[0162] Next, for each paragraph, the prompt creation unit 135 places all of the sentences in the paragraph in the <inquiry information> of the prompt template information paired with the class condition: "Please tell me with YES / NO whether the number of characters in the following sentence group is 500 characters or less. Sentence group "<inquiry information>"", and creates a prompt. The prompt creation unit 135 creates, for example, "Please tell me with YES / NO whether the number of characters in the following sentence group is 500 characters or less. Sentence group "The information processing device of the first invention is an information processing device comprising... With this configuration, it is possible to use the generation AI by using an appropriate prompt according to the class of the inquiry information. Furthermore, the information processing device of the second invention is an information processing device that is... compared to the first invention. With this configuration, it is possible to use the generation AI by using an appropriate prompt according to the class of the information of the invention, and to support the creation of patent documents."

[0163] Next, the answer acquisition unit 136 passes each created prompt to the generation AI. Then, the answer acquisition unit 136 acquires an answer ("YES" or "NO") from the generation AI for each prompt. Note that the answer to the inquiry information "The information processing device of the first invention is an information processing device comprising... With such a configuration, the generation AI can be used by using an appropriate prompt according to the class of the inquiry information. Furthermore, the information processing device of the second invention is an information processing device that is... compared to the first invention. With such a configuration, the generation AI can be used to support the creation of patent documents by using an appropriate prompt according to the class of the information of the invention." was "NO."

[0164] Furthermore, the query acquisition unit 131 acquires components from a sentence corresponding to the tag [Form for carrying out the invention] in the file "Specification20231125.docx" based on the class condition "<tag>Form for carrying out the invention <linguistic characteristics>Component" for "ID=2." Here, the query acquisition unit 131 acquires, as components, noun phrases (reception unit, query acquisition unit, terminal storage unit, etc.) immediately followed by a code (numeric string).

[0165] Next, the prompt creation unit 135 creates a prompt by placing each component in the <inquiry information> of the prompt template information paired with the class condition, "Please list terms that are subordinate to the <inquiry information>!" For example, the prompt creation unit 135 creates "Please list terms that are subordinate to the reception section!"

[0166] Next, the answer acquisition unit 136 passes each created prompt to the generation AI. Then, the answer acquisition unit 136 acquires an answer from the generation AI for each prompt. It should be noted that the answer to the inquiry information "reception department" is "provide information about the sub-concept of the general 'reception department'. The sub-concept of the reception department may include the following functions and elements: reception desk, information counter, guest management, visitor registration, appointment adjustment, security check-in, telephone answering, and guidance service."

[0167] Furthermore, the inquiry acquisition unit 131 acquires technical terms from a sentence corresponding to the tag [Form for carrying out the invention] in the file "Specification 20231125.docx" based on the class condition "<tag>Form for carrying out the invention <linguistic characteristics>Technical terminology" for "ID=2." Note that a technique for acquiring technical terms from a sentence can be realized by a publicly known technique such as "termextract" (URL: http: / / gensen.dl.itc.u-tokyo.ac.jp / pytermextract / ). Here, the inquiry acquisition unit 131 acquires the technical term "generation AI" from, for example, the sentence "The inquiry acquisition unit 131 acquires inquiry information, which is information included in the reception information and is information used to query the generation AI," in the tag [Form for carrying out the invention].

[0168] Next, the prompt creation unit 135 creates a prompt by placing each technical term in the <inquiry information> of the prompt template information paired with the class condition, "Please tell me the definition and specific example of <inquiry information>!" For example, the prompt creation unit 135 creates "Please tell me the definition and specific example of generated AI!"

[0169] Next, the answer acquisition unit 136 passes each created prompt to the generation AI. Then, the answer acquisition unit 136 acquires an answer from the generation AI for each prompt. It should be noted that the answer to the inquiry information "generation AI" was "Generation AI is a type of artificial intelligence that refers to a system that generates new content based on input data. Specific examples include: Text generation: AI such as the GPT series can create stories or articles from given prompts. Image generation: DALL-E and this AI system can.... Music generation: AI such as AIVA can.... Speech synthesis: It can...."

[0170] Furthermore, the query acquisition unit 131 sequentially acquires sentences from the file "specification20231125.docx" based on the class condition "<tag>-<linguistic property> sentence" of "ID=4." Each sentence is query information.

[0171] Next, the prompt creation unit 135 places each sentence in the <inquiry information> section of the prompt template information "If the following sentence is a compound sentence, change it into multiple simple sentences with a subject! Sentence "<inquiry information>"". As a result, the prompt creation unit 135 creates, for example, a prompt such as "If the following sentence is a compound sentence, change it into multiple simple sentences with a subject! Sentence "The terminal storage unit 21 is preferably a non-volatile recording medium, but can also be realized with a volatile recording medium, and the process by which information is stored in the terminal storage unit 21 does not matter."

[0172] Next, the answer acquisition unit 136 passes each created prompt to the generation AI. Then, the answer acquisition unit 136 acquires an answer from the generation AI for each prompt. Note that the answer to the inquiry information "The terminal storage unit 21 is preferably a non-volatile recording medium, but can also be realized as a volatile recording medium, and the process by which information is stored in the terminal storage unit 21 is not important" was "The terminal storage unit 21 is preferably a non-volatile recording medium, but can also be realized as a volatile recording medium. The process by which information is stored in the terminal storage unit 21 is not important."

[0173] Next, the output acquisition unit 137 acquires output information as follows: That is, the output acquisition unit 137 acquires, from the template management table, the process types paired with the class conditions of "ID=2" and "ID=4" in the template management table.

[0174] Next, the output acquisition unit 137 adds the response corresponding to the processing type "check" to the received information "specification 20231125.docx" as a comment to the inquiry information. If the response to the prompt template information "Please tell us with YES / NO whether the following sentence has a subject or not. <inquiry information>" is "NO", the output acquisition unit 137 adds "This sentence does not have a subject" as a comment to the inquiry information (sentence). An example of such a comment is 1101 in FIG. 11.

[0175] Furthermore, if the answer to the prompt template information "Please tell us with a YES / NO whether the number of characters in the following sentence group is 500 characters or less. Sentence group "<Inquiry information>"" is "NO," the output acquisition unit 137 adds "This paragraph is longer than 500 characters" as a comment to the inquiry information (paragraph). An example of such a comment is 1102 in FIG. 11.

[0176] Furthermore, the output acquiring unit 137 places a response corresponding to the processing type "add" immediately after the sentence that is the query information, the paragraph that is the query information, or a sentence that includes the query information. For example, a response to the query information "reception unit" is placed immediately after the sentence "The reception unit 12 receives reception information from the user." that includes the query information "reception unit." Such a response is 1103 in FIG. 11.

[0177] Furthermore, the output acquisition unit 137 places, for example, a response to the inquiry information "Generation AI" immediately after a sentence including the generation AI: "The inquiry acquisition unit 131 acquires inquiry information, which is information included in the received information and is information used to inquire to the generation AI." Such a response is 1104 in FIG.

[0178] Furthermore, the output acquisition unit 137 replaces the answer corresponding to the processing type "change" with the inquiry information. For example, the output acquisition unit 137 replaces the inquiry information, which is a complex sentence, "The terminal storage unit 21 is preferably a non-volatile recording medium, but can also be realized as a volatile recording medium, and the process by which information is stored in the terminal storage unit 21 does not matter" with the answer to the inquiry information, "The terminal storage unit 21 is preferably a non-volatile recording medium, but can also be realized as a volatile recording medium. The process by which information is stored in the terminal storage unit 21 does not matter."

[0179] Next, the output unit 14 outputs the revised specification "20231125.docx" using the answer of the generation AI. An example of such output is shown in FIG.

[0180] (Specific Example 3) It is assumed that the user inputs into the information processing device 1 a file "abstract 20231125.docx" that stores the character string shown in FIG.

[0181] Next, the reception unit 12 of the information processing device 1 receives the reception information "abstract20231125.docx" from the user. Next, the inquiry acquisition unit 131 determines that the reception information is not a "question" based on the contents of the received file "abstract20231125.docx."

[0182] Next, the inquiry acquisition unit 131 etc. processes the received reception information as follows, and outputs output information.

[0183] First, the class determination unit 132 obtains the character string "abstract" paired with the tag [document name] in the file of reception information "abstract20231125.docx", and obtains the document type "abstract" from the character string.

[0184] Next, the class determination unit 132 acquires the class conditions for "ID=3" and "ID=4" that match the type "abstract." For convenience of explanation, the following description will be given for the case where only the class condition for "ID=3" is applied.

[0185] Next, the class determination unit 132 obtains the class condition "<tag>-<linguistic property>-" for "ID=3" from the template management table (FIG. 7). The class determination unit 132 also obtains the class condition "<tag>-<linguistic property> sentence" for "ID=4" from the template management table (FIG. 7).

[0186] The template acquisition unit 134 also acquires prompt template information paired with the class condition "ID=3" from the template management table (FIG. 7) in association with each class condition.

[0187] Next, the inquiry acquisition unit 131 acquires the received information (the entire abstract in FIG. 12) as inquiry information from the class condition "<tag>-<linguistic characteristic>-".

[0188] Next, the prompt creation unit 135 creates a prompt by replacing the <inquiry information> in the prompt template information "If the following document is longer than 400 characters, summarize it in 400 characters or less!" with the entire abstract (Figure 12).

[0189] Next, the answer acquisition unit 136 passes each created prompt to the generation AI, and the answer acquisition unit 136 then obtains a summary of up to 400 characters.

[0190] Next, the output acquisition unit 137 acquires output information as follows: That is, the output acquisition unit 137 acquires the process type "change" that pairs with the class condition of "ID=3" in the template management table from the template management table.

[0191] Furthermore, for a response corresponding to the processing type "change," the output acquisition unit 137 replaces the inquiry information with the response. That is, the output acquisition unit 137 acquires the acquired response as a summary.

[0192] Next, the output unit 14 outputs the abstract obtained using the answer of the generation AI. An example of such output is shown in FIG.

[0193] (Example 4) It is assumed that the user (Yamada A) inputs information for a sales report into each field of the input screen (FIG. 14) displayed on the terminal device 2 and presses the "Register" button.

[0194] Next, terminal device 2 accepts Yamada A's input and uses the collection of information entered in each field to compose information to be sent to the information processing device. Such information is, for example, as shown in Figure 15. Next, terminal device 2 sends the information (Figure 15) to information processing device 1.

[0195] Next, the reception unit 12 of the information processing device 1 receives the reception information of FIG.

[0196] Next, the class determination unit 132 acquires the document type “sales daily report” from “<document> sales daily report” included in the reception information received by the reception unit 12 .

[0197] Next, the class determination unit 132 acquires the class condition for "ID=51" that matches the type "sales daily report." Note that, here, the class condition for "ID=4" is usually adopted, but as described above, the processing for the class condition for "ID=4" is not described.

[0198] Next, the class determination unit 132 obtains the class conditions for "ID=51", "<tag> departure station, arrival seat <linguistic characteristics>-" and "<tag> sales report <linguistic characteristics>-", from the template management table (Figure 7).

[0199] The template acquisition unit 134 also acquires prompt template information "Tell me the travel fare from <Departure station> to <Arrival station>!" that is paired with the class condition of "ID=51" from the template management table (FIG. 7).

[0200] The template acquisition unit 134 also acquires, from the template management table (FIG. 7), another prompt template information paired with the class condition of "ID=51": "Summarize the following sales report in 400 characters or less! [Sales report] <Sales report>."

[0201] Next, the inquiry acquisition unit 131 acquires the inquiry information "<Departure station> Osaka, <Arrival station> Tokyo" corresponding to the tags <Departure station> and <Arrival station> in the class condition from the reception information (Figure 15).

[0202] In addition, the inquiry acquisition unit 131 acquires inquiry information corresponding to the tag <sales report> in the class condition, such as "<sales report> Today, I visited Mr. XXX of XX Co., Ltd. in Tokyo..." from the reception information (Figure 15).

[0203] Next, the prompt creation unit 135 substitutes each of the inquiry information "<Departure station> Osaka, <Arrival station> Tokyo" into the variable portion of the prompt template information "Please tell me the travel cost from <Departure station> to <Arrival station>!" to create the first prompt "Please tell me the travel cost from Osaka to Tokyo!".

[0204] In addition, the prompt creation unit 135 substitutes the inquiry information "<Sales report> Today, I visited Mr. / Ms. XXX of XX Co., Ltd. in Tokyo..." into the variable portion of the prompt template information "Summarize the following sales report in 400 characters or less! [Sales report] <Sales report>" to create a second prompt "Summarize the following sales report in 400 characters or less! [Sales report] Today, I visited Mr. / Ms. XXX of XX Co., Ltd. in Tokyo...".

[0205] Next, the answer acquisition unit 136 passes the created first prompt to the generation AI, and then the answer acquisition unit 136 acquires the answer "14,720 yen" from the generation AI.

[0206] The answer acquisition unit 136 also passes the created second prompt to the generation AI. The answer acquisition unit 136 then acquires an answer from the generation AI: "Today, I visited Mr. XXX of XXXX Corporation in Tokyo and introduced him to our AI-based daily sales report system. He was particularly interested in the generation AI function, and..."

[0207] Next, the output acquisition unit 137 uses the response "14,720 yen" from the generation AI and the processing type (insert (transportation expenses, expense DB)) to construct the output information "insert (14,720 yen, expense DB)." Next, the output unit 14 executes "insert (14,720 yen, expense DB)" and inputs the transportation expenses of "14,720 yen" into the expense DB (not shown).

[0208] Furthermore, the output acquisition unit 137 acquires the response from the generation AI, "Today, I visited Mr. XXX of XX Co., Ltd. in Tokyo and introduced him to our sales daily report system using AI. He was particularly interested in the generation AI function, ..." as a processing type (send("Today, I visited Mr. XXX of XX Co., Ltd. in Tokyo and introduced him to our sales daily report system using AI. He was particularly interested in the generation AI function, ...", $supervisor's address)) to acquire the output information "send("Today, I visited Mr. XXX of XX Co., Ltd. in Tokyo and introduced him to our sales daily report system using AI. He was particularly interested in the generation AI function, ...", Yamada A's supervisor's address)". Next, the output unit 14 executes the output information. As a result, a summary of Yamada A's sales report is displayed on Yamada A's supervisor's

[0209] As described above, according to this embodiment, an appropriate prompt is created according to the class of inquiry information, and the generated AI can be used using the prompt.

[0210] Furthermore, according to this embodiment, an appropriate prompt can be created according to the class of invention information, and the prompt can be used to utilize generation AI to assist in the creation of patent documents.

[0211] Furthermore, according to this embodiment, it is possible to provide a generative AI control engine that makes the most of generative AI.

[0212] The processing in this embodiment may be realized by software. This software may be distributed by software download or the like. This software may also be recorded on a recording medium such as a CD-ROM and distributed. This also applies to the other embodiments in this specification. The software that realizes information system A in this embodiment is the following program. In other words, this program is a program for causing a computer that can access a template management unit in which prompt template information is stored in correspondence with one or more class conditions, which are conditions related to the class of a string, to function as: a reception unit that receives reception information from a user; a query acquisition unit that acquires query information, which is information contained in the reception information and is information used to query a generation AI; a template acquisition unit that acquires from the template management unit prompt template information corresponding to the class conditions satisfied by the class of the query information acquired by the query acquisition unit; a prompt creation unit that creates a prompt using the prompt template information acquired by the template acquisition unit and the query information acquired by the query acquisition unit; an answer acquisition unit that passes the prompt created by the prompt creation unit to the generation AI and acquires an answer from the generation AI; an output acquisition unit that acquires output information using the answer acquired by the answer acquisition unit; and an output unit that outputs the output information.

[0213] 16 shows the appearance of a computer that executes the programs described herein to realize the information processing device 1 and the like according to the various embodiments described above. The above-described embodiments can be realized by computer hardware and a computer program executed thereon. FIG. 16 is an overview of this computer system 300, and FIG. 17 is a block diagram of the system 300.

[0214] In FIG. 16, a computer system 300 includes a computer 301 including a CD-ROM drive, a keyboard 302, a mouse 303, and a monitor 304.

[0215] 17, the computer 301 includes, in addition to a CD-ROM drive 3012, an MPU 3013, a bus 3014 connected to the CD-ROM drive 3012 etc., a ROM 3015 for storing programs such as a boot-up program, a RAM 3016 connected to the MPU 3013 for temporarily storing instructions for application programs and providing temporary storage space, and a hard disk 3017 for storing application programs, system programs, and data. Although not shown here, the computer 301 may further include a network card for providing connection to a LAN.

[0216] A program that causes computer system 300 to execute the functions of information processing device 1 and the like of the above-described embodiment may be stored on CD-ROM 3101, inserted into CD-ROM drive 3012, and then transferred to hard disk 3017. Alternatively, the program may be transmitted to computer 301 via a network (not shown) and stored on hard disk 3017. The program is loaded into RAM 3016 when executed. The program may also be loaded directly from CD-ROM 3101 or the network.

[0217] The program does not necessarily have to include an operating system (OS) or a third-party program that causes the computer 301 to execute the functions of the information processing device 1 of the above-described embodiment. The program only needs to include instructions that call appropriate functions (modules) in a controlled manner and achieve the desired results. How the computer system 300 operates is well known, and a detailed description thereof will be omitted.

[0218] In addition, in the above program, the steps of transmitting information and receiving information do not include processing performed by hardware, such as processing performed by a modem or interface card in the transmission step (processing that can only be performed by hardware).

[0219] The computer that executes the program may be a single computer or a plurality of computers, that is, it may perform centralized processing or distributed processing.

[0220] Furthermore, in each of the above embodiments, it goes without saying that two or more communication means present in one device may be physically realized by one medium.

[0221] Furthermore, in each of the above embodiments, each process may be realized by centralized processing in a single device, or may be realized by distributed processing in a plurality of devices.

[0222] The present invention is not limited to the above-described embodiment, and various modifications are possible, and it goes without saying that these modifications are also included within the scope of the present invention.

[0223] As described above, the information processing device 1 of the present invention has the advantage of being able to utilize generation AI using appropriate prompts according to the class of inquiry information, and is useful as a specification creation support device, document creation support device, etc.

Claims

1. An information processing device comprising: a template management unit in which prompt template information is stored in correspondence with two or more class conditions, the condition being related to a class of strings, and including one or two of the following conditions: a condition for a tag in the reception information or a condition for a linguistic characteristic of a string in the reception information; a reception unit for receiving reception information; a query acquisition unit for acquiring query information, the information included in the reception information, a string that matches each of the two or more class conditions, and a string to be used for querying a generation AI; a template acquisition unit for acquiring from the template management unit prompt template information that pairs with the class condition matched by the query information acquired by the query acquisition unit; a prompt creation unit for creating a prompt using the prompt template information acquired by the template acquisition unit and the query information acquired by the query acquisition unit; an answer acquisition unit for passing the prompt created by the prompt creation unit to a generation AI and acquiring an answer from the generation AI; an output acquisition unit for acquiring output information using the answer acquired by the answer acquisition unit; and an output unit for outputting the output information.

2. An information processing apparatus according to claim 1, wherein the received information is information relating to an invention, and the output information is information constituting a patent specification, claims, or abstract.

3. The information processing device of claim 1, wherein the inquiry acquisition unit acquires two or more pieces of inquiry information from the reception information for at least one class condition, and the prompt creation unit creates a prompt by placing each of the two or more pieces of inquiry information for the one class condition in prompt template information acquired for the one class condition.

4. An information processing device according to claim 1, wherein the output acquisition unit acquires the output information by arranging a part or all of the answers in output template information, which is a template for constructing output information.

5. An information processing device according to claim 1, wherein the linguistic characteristics are "sentences," "paragraphs," "words," "constituent elements" that identify an invention, or "technical terms." 6. An information processing device as described in claim 1, further comprising a class determination unit which determines whether the reception information is a "question", wherein when the class determination unit determines that the reception information is a "question", the template acquisition unit does not acquire the prompt template information, and the prompt creation unit creates a prompt including the reception information.

7. The information processing apparatus according to claim 1, wherein the reception information is a document, and the class includes a type of the document.

8. An information processing device as described in claim 1, further comprising a generation AI determination unit that determines a generation AI from two or more generation AIs in accordance with the prompt template information acquired by the template acquisition unit, wherein the answer acquisition unit passes the prompt created by the prompt creation unit to the generation AI determined by the generation AI determination unit and acquires an answer from the generation AI.

9. An information processing device as described in claim 1, wherein the class is specified by a field identifier that identifies one or more fields on a screen, the class condition is a condition relating to one or more field identifiers, the reception unit receives the reception information entered into one or more screen fields, and the inquiry acquisition unit acquires the inquiry information based on the information entered into one or more fields.

10. An information processing device as described in claim 1, wherein each of the two or more prompt template information of the template management unit is associated with a processing type that specifies the type of processing for the output information, the processing types including "add" and "change", and the output acquisition unit, when the processing type paired with the prompt template information used to obtain the answer obtained by the answer acquisition unit is "add", places the answer or a part of the answer at a predetermined position with respect to the inquiry information that is the source of the answer, and when the processing type paired with the prompt template information used to obtain the answer obtained by the answer acquisition unit is "change", updates the inquiry information that is the source of the answer to the answer or the part of the answer.

11. An information processing method realized by a template management unit in which prompt template information is stored in association with two or more class conditions, the two or more class conditions being conditions related to a class of character strings, the two or more class conditions including one or two conditions of a tag condition in the received information or a linguistic characteristic condition of a character string in the received information, a reception unit, a query acquisition unit, a template acquisition unit, a prompt creation unit, a response acquisition unit, an output acquisition unit, and an output unit, comprising: a reception step in which the reception unit receives the received information; a query acquisition step in which the query acquisition unit acquires query information, the query information being information included in the received information, a string that matches each of the two or more class conditions, and a string to be used in a query to a generation AI; a template acquisition step in which the template acquisition unit acquires, from the template management unit, prompt template information corresponding to a class condition satisfied by the class of the query information acquired by the query acquisition unit; and a prompt creation step in which the prompt creation unit creates a prompt using the prompt template information acquired by the template acquisition unit and the query information acquired by the query acquisition unit. An information processing method comprising: an answer acquisition step in which the answer acquisition unit passes the prompt created by the prompt creation unit to a generation AI and acquires an answer from the generation AI; an output information acquisition step in which the output acquisition unit acquires output information using the answer acquired by the answer acquisition unit; and an output step in which the output unit outputs the output information.

12. A recording medium having recorded thereon a computer capable of accessing a template management unit in which prompt template information is stored in correspondence with two or more class conditions, the conditions being conditions related to a class of strings, including one or two of the following conditions: conditions for a tag in the reception information or conditions for the linguistic characteristics of a string in the reception information; the recording medium having recorded thereon a program for causing the recording medium to function as an output unit that outputs the output information: a reception unit that receives reception information; a query acquisition unit that acquires query information, the information included in the reception information, which is a string that matches each of the two or more class conditions and is used to query a generation AI; a template acquisition unit that acquires from the template management unit prompt template information corresponding to the class conditions satisfied by the class of the query information acquired by the query acquisition unit; a prompt creation unit that creates a prompt using the prompt template information acquired by the template acquisition unit and the query information acquired by the query acquisition unit; an answer acquisition unit that passes the prompt created by the prompt creation unit to a generation AI and acquires an answer from the generation AI; an output acquisition unit that acquires output information using the answer acquired by the answer acquisition unit; and

Images